﻿* Encoding: UTF-8.
***analyzing pew 2012 data***

***Important variables***
•	Q27  a thru e – reasons for political unfriending 
•	Q16 talk politics 
•	Q20 a thru d – political reasons for using sns 
•	Q21 – how much of what you post is political 
•	Q22 – how much your friends post political 
•	Q23 – agree/disagree with pol posts 
•	Age
•	Educ
•	Hisp
•	Race
•	Party + partyln
•	Ideo 
•	Inc 

FREQUENCIES q27a q27b q27c q27d q27e.

corr q27a q27b q27c q27d q27e.

**posting too frequently - 9.7%**

recode q27a (1=1) (2=0) (8=sysmis) (9=sysmis) into ufptm. 
FREQUENCIES ufptm.

**disagreed or found offensive - 9.3%**

FREQUENCIES q27b.
recode q27b (1=1) (2=0) (8=sysmis) (9=sysmis) into ufdo. 
FREQUENCIES ufdo.

**argued - 7.8%**

FREQUENCIES q27c. 
recode q27c (1=1) (2=0) (8=sysmis) (9=sysmis) into ufa. 
FREQUENCIES ufa.

**disagreed with your polpost - 4.3%**

FREQUENCIES q27d. 
recode q27d (1=1) (2=0) (8=sysmis) (9=sysmis) into ufdy. 
FREQUENCIES ufdy.

**worried they would offend your friends 5.4% **

FREQUENCIES q27e. 
recode q27e (1=1) (2=0) (8=sysmis) (9=sysmis) into ufwo. 
FREQUENCIES ufwo.

**alpha .77**

RELIABILITY
  /VARIABLES=ufptm ufdo ufa ufdy ufwo
  /FORMAT=NOLABELS
  /SCALE(ALPHA)=ALL/MODEL=ALPHA
  /STATISTICS=DESCRIPTIVE SCALE CORR
  /SUMMARY=TOTAL. 

**mean .07 sd .19**

compute unfriendscale = mean (ufptm, ufdo, ufa, ufdy, ufwo).
FREQUENCIES unfriendscale.
DESCRIPTIVES unfriendscale. 

FREQUENCIES q16.
recode q16 (1=4) (2=3) (3=2) (4=1) (8=sysmis) (9=sysmis) into talkpol.
FREQUENCIES talkpol.
DESCRIPTIVES talkpol.

FREQUENCIES q20a q20b q20c q20d.

**political news**

FREQUENCIES q20a.
recode q20a (1=3) (2=2) (3=1) (4=0) (5=0) (8=sysmis) (9=sysmis) into polnews.
FREQUENCIES polnews. 

**discussing politics**

FREQUENCIES q20b.
recode q20b (1=3) (2=2) (3=1) (4=0) (5=0) (8=sysmis) (9=sysmis) into poldiscuss.
FREQUENCIES poldiscuss.

*finding others who share your views*

FREQUENCIES q20c.
recode q20c (1=3) (2=2) (3=1) (4=0) (5=0) (8=sysmis) (9=sysmis) into polshare.
FREQUENCIES polshare.

**recruiting others to get involved8*

FREQUENCIES q20d.
recode q20d (1=3) (2=2) (3=1) (4=0) (5=0) (8=sysmis) (9=sysmis) into polrecruit.
FREQUENCIES polrecruit.

**alpha .87**

RELIABILITY
  /VARIABLES=polnews poldiscuss polshare polrecruit 
  /FORMAT=NOLABELS
  /SCALE(ALPHA)=ALL/MODEL=ALPHA
  /STATISTICS=DESCRIPTIVE SCALE CORR
  /SUMMARY=TOTAL. 

corr polnews poldiscuss polshare polrecruit.

**0 to 3 mean = .88 sd = .84**

compute polfbmotiv = mean(polnews, poldiscuss, polshare, polrecruit). 
FREQUENCIES polfbmotiv. 
DESCRIPTIVES polfbmotiv.

**how much you post is political **

FREQUENCIES q21.
recode q21 (1=4) (2=3) (3=2) (4=1) (5=0) (8=sysmis) (9=sysmis) into postpol.
FREQUENCIES postpol.

**how much others post is political **

FREQUENCIES q22. 
recode q22 (1=4) (2=3) (3=2) (4=1) (5=0) (8=sysmis) (9=sysmis) into friendspostpol.
FREQUENCIES friendspostpol.
DESCRIPTIVES friendspostpol.

**how much you agree/ disagree**

FREQUENCIES q23a. 
recode q23a (1=3) (2=2) (3=1) (4=0) (8=sysmis) (9=sysmis) into agree.
FREQUENCIES agree.

 FREQUENCIES q23b.
recode q23b (1=3) (2=2) (3=1) (4=0) (8=sysmis) (9=sysmis) into disagree.
FREQUENCIES disagree.
DESCRIPTIVES disagree.

**sign but only -.096**

corr agree disagree. 

*mean 46.4 sd 18.3**

FREQUENCIES age. 
DESCRIPTIVES age. 
recode age (98=sysmis) (99=sysmis).
FREQUENCIES age. 

*mean 4.3 sd 1.7 category 4 is technical, trade, or vocational school**

FREQUENCIES educ.
DESCRIPTIVES educ. 
recode educ (8=sysmis) (9=sysmis).
DESCRIPTIVES educ. 

**14.2% hispanic**

FREQUENCIES hisp.
recode hisp (8=sysmis) (9=sysmis).
recode hisp (1=1) (2=0). 
DESCRIPTIVES hisp.
FREQUENCIES hisp.

**23.8% nonwhite**

FREQUENCIES race.
recode race (1=0) (8=sysmis) (9=sysmis) (else=1) into nonwhite. 
FREQUENCIES nonwhite. 

**original 1 R 2 d 3 ind **
**NEW 1R 27.2 2I 37.2 3D 35.6**

FREQUENCIES party. 
recode party (1=1) (2=3) (3=2) (else=sysmis) into party3. 
FREQUENCIES party3.

**higher is more liberal**
**mean 2.88 sd 1.03**

FREQUENCIES ideo.
recode ideo (1=1) (2=2) (3=3) (4=4) (5=5) (else=sysmis) into ideo5.
FREQUENCIES ideo5.
DESCRIPTIVES ideo5.

recode ideo5 (1=2) (2=1) (3=0) (4=1) (5=2) into ideostrength.
FREQUENCIES ideostrength.
DESCRIPTIVES ideostrength.


**mean 4.74 sd 2.46 category 4 is 30-40K**

FREQUENCIES inc.

recode inc (1=1) (2=2) (3=3) (4=4) (5=5) (6=6) (7=7) (8=8) (9=9) (else=sysmis) into statedinc.
FREQUENCIES statedinc. 
DESCRIPTIVES statedinc. 

**male is 1, female is 0*
51.3%female

FREQUENCIES sex.
recode sex (1=1) (2=0) into gender. 
FREQUENCIES gender. 

*STRONG corr at .6**

corr polfbmotiv postpol.

*****ANALYSIS****
**overall political unfriending**

REGRESSION
  /MISSING LISTWISE
  /STATISTICS COEFF OUTS R ANOVA
  /CRITERIA=PIN(.05) POUT(.10)
  /NOORIGIN 
  /DEPENDENT unfriendscale
  /METHOD=ENTER  gender age educ hisp nonwhite statedinc party3 ideostrength
  /METHOD=ENTER  talkpol polfbmotiv friendspostpol disagree.

**more ideological (both sides) more likely to unfriend**

means unfriendscale by ideo5.

**Breaking it down by individual types of unfriending**
ufptm ufdo ufa ufdy ufwo

LOGISTIC REGRESSION VARIABLES ufptm
  /METHOD=ENTER  gender age educ hisp nonwhite statedinc party3 ideo5
   /METHOD=ENTER  talkpol polfbmotiv postpol friendspostpol disagree
  /CRITERIA=PIN(.05) POUT(.10) ITERATE(20) CUT(.5).

LOGISTIC REGRESSION VARIABLES ufdo
  /METHOD=ENTER  gender age educ hisp nonwhite statedinc party3 ideo5
   /METHOD=ENTER  talkpol polfbmotiv postpol friendspostpol disagree
  /CRITERIA=PIN(.05) POUT(.10) ITERATE(20) CUT(.5).

LOGISTIC REGRESSION VARIABLES ufa
  /METHOD=ENTER  gender age educ hisp nonwhite statedinc party3 ideo5
   /METHOD=ENTER  talkpol polfbmotiv postpol friendspostpol disagree
  /CRITERIA=PIN(.05) POUT(.10) ITERATE(20) CUT(.5).

LOGISTIC REGRESSION VARIABLES ufdy
  /METHOD=ENTER  gender age educ hisp nonwhite statedinc party3 ideo5
   /METHOD=ENTER  talkpol polfbmotiv postpol friendspostpol disagree
  /CRITERIA=PIN(.05) POUT(.10) ITERATE(20) CUT(.5).

LOGISTIC REGRESSION VARIABLES ufwo
  /METHOD=ENTER  gender age educ hisp nonwhite statedinc party3 ideo5
   /METHOD=ENTER  talkpol polfbmotiv postpol friendspostpol disagree
  /CRITERIA=PIN(.05) POUT(.10) ITERATE(20) CUT(.5).







